I recently attended the Digital Humanities section of the annual Canadian academic Congress. Ian Lancashire of the University of Toronto was on my panel and he gave a very moving paper on the correlation between language decline in writers and the onset of alzheimers. His most famous paper on this topic concerns Agatha Christie — Lancashire could identify when her disease set in by noticing the decreasing diversity of her vocabulary after a certain date. Now he is talking about two more writers who were identified with alzheimer’s, one a detective fiction writer, another a children’s books writer.
Lancashire’s findings are potentially interesting not just as a historical curiosity, but because they could lead to some sort of personal screening device. We all know that by the time a disease is diagnosed it is often too late. In Lancashire’s case, his identification of language decline in his writers always predates the diagnosis of the disease in their lives. As he surmised during the Q&A, we might be able to create a text-tracking system on our computers to monitor our emails over time to detect significant diminishment of our personal vocabularies. That way, doctors would have a tool to try to address our mental health via our writing. Now that we all produce so much writing — it’s not just the province of novelists anymore — this is something that could be used by many people.
You can imagine different scenarios — your computer tells you to go to the doctor because it thinks you have alzheimer’s, but in actuality you’ve just become a very bad email writer, or are going through an emotionally difficult period, or maybe you just email too often when you’re drunk……Or it is right and it helps you get treatment earlier. Either way it is at the forefront of personal computational diagnostics, intrusive and useful at the same time.
What an example of the way text mining can solve very real, very personal problems.